A Spectral-Based Approach for BCG Signal Content Classification

This paper has two objectives: the first is to generate binary flags indicate useful frames permitting measurement of cardiac and respiratory rates from Ballistocardiogram (BCG) signals—in fact, human body activities during measurements can disturb BCG signal content, leading difficulties in vital sign measurement; second objective achieve refined segmentation according these activities. The proposed framework makes use approaches: an unsupervised classification based on Gaussian Mixture Model (GMM) a supervised K-Nearest Neighbors (KNN). Both approaches consider spectral features, namely Spectral Flatness Measure (SFM) Centroid (SC), determined feature extraction step. Unsupervised used explore content signals, justifying existence different classes definition hyper-parameters for effective segmentation. In contrast, considered approach aims determine if allows heart rate (HR) (RR) or not. Furthermore, levels are classify human-body into many realistic (e.g., coughing, holding breath, air expiration, movement, et al.). one considers frame-by-frame classification, while one, aiming boost performance, transforms SFM SC features temporal series which track variation measures signal. constitutes novelty this field represents powerful method segment signals activities, resulting accuracy 94.6%.